Meta Ads Case Study: 1,083 Purchases at 4.06 ROAS for Hummingbird

Ad examples

Bundle and first-use offers for the Hummingbird head shaver. Auto-scrolls - hover to pause.

The situation

Hummingbird sells a head shaver at humbirdusa.com. Shoppers can already buy a basic shave at Walmart or another retailer. An ad that only argues price does not give a reason to switch. The first job was a first use, not a sticker comparison.

The Google Ads export for the later window is the company account Sean identified as Top Technologies. Facebook is where this case starts. The early months include the holiday push. This page does not itemize that calendar. It reports the purchases on the export and the offer that made those purchases possible. The brand's read of the window is that the purchase volume was the result they wanted.

The insight

The offer had to be a package. Bundles, and a discount aimed at a new buyer's first order, gave the product a reason to be chosen. Promotions were tested until one bundle and one discount held. Budget moved after that test, not before it.

Top of funnel carried the early account. Bottom of funnel was the weaker layer. The discount pulled impulsive orders and a short shopping trip, which is why that layer did not look as strong. Changing that funnel came later. This page stays on the early Facebook result.

What we deployed

Bundles and a first-use discount Packages replaced a single-unit price pitch. The discount was there to get a new buyer to try the shaver once, not to match a retailer sticker.

Test the promotion, then scale it Offer tests stayed small until the bundle and the discount were clear. Once that combination held, spend moved onto it and the account scaled.

Facebook first, Google during the holiday The account started on Facebook. Google Ads came on during the holiday period, after the offer was already working on Meta. The Google export on file is the 2025 account view, not the Facebook crop.

Top of funnel as the working layer Purchases in the export sit on the active top-of-funnel row. Bottom of funnel stayed on, with a higher frequency and a lower purchase ROAS. This page does not walk through the later funnel change.

Results

The Facebook export is the file labeled October. The crop does not print a date range. Filled rows only:

One active top-of-funnel campaign: frequency 1.86, Purchase ROAS 4.06, 1,083 purchases, CPM $6.04.

One active bottom-of-funnel campaign: frequency 4.78, Purchase ROAS 2.63, 179 purchases, CPM $6.00.

A second active top-of-funnel row shows 104 purchases. Its ROAS cell is blank. CPM on that row is $6.13.

The result row prints frequency 2.46 and CPM $6.05. The account Purchase ROAS cell and the purchase total cell are blank, so this page does not invent a summed ROAS from the rows.

The working target was about a 2 ROAS. The team's read of the scaled month is about 3.4 ROAS, with November spend past $10,000. Neither figure is printed on this export.

Google Ads, Jan 31 - Aug 18, 2025, 13 campaigns: 5.18K clicks, 60.00 conversions, $4.16K cost, $0.80 average CPC. That is the later search account, not the Facebook crop. The brand was satisfied with the purchase count on the Meta export.

Why it worked

Retail price was not an offer A Walmart sticker wins a price fight. The bundle gave a first-use reason that the shelf price did not.

The test had a finish line Spend waited until the promotion and the bundle were clear, then it moved. Scale followed the test, including the holiday month, without a new offer invented for that week.

Top of funnel matched the buy The 4.06 ROAS and 1,083 purchases sit on top of funnel. Bottom of funnel's 2.63 ROAS and higher frequency match a shorter, discount-led trip. That is the early account. The later funnel edit is not part of this result.

← All case studies · Related service →